Trillions of dollars in investment, thousands of new players – AI data centers have become a "new gold rush" that concerns the fate of the global economy

Wallstreetcn
2025.12.22 03:57
portai
I'm PortAI, I can summarize articles.

The global construction frenzy of AI data centers is sweeping the world, with thousands of "new players"—energy companies, Bitcoin mining firms, and private equity funds—rushing into this multi-trillion-dollar gamble. Tech giants cleverly shift risks through leasing strategies, offloading debt pressure onto inexperienced developers. However, project delays and contract terminations are frequent, intensifying market concerns about the AI bubble. Once the AI bubble arrives, risks may sweep across the globe

A gold rush driven by artificial intelligence (AI) is reshaping the global data center market, with the main players no longer just tech giants, but thousands of new entrants from diverse backgrounds. This wave is attracting trillions of dollars in investment and could have far-reaching impacts on the global economy.

According to data compiled by Bloomberg News, the scale of this construction boom is unprecedented. By 2025, the credit transaction volume for data centers in the United States is expected to reach at least $178.5 billion. Tech giants like Oracle, Meta, and Alphabet have helped push this year's global bond issuance total to over $6.57 trillion. JP Morgan predicts that issuers will tap nearly all major debt markets to fund this global construction.

However, the core change of this boom lies in the composition of market participants. Data shows that the majority of future planned data center capacity will be built by "new players" and other industry participants outside of large tech companies. This structural shift not only diversifies risk but also brings new concerns: if the commercial prospects of AI are disproven, its collapse could ripple through every corner of the global economy, affecting every aspect of equity and debt markets.

Large tech companies are also adjusting their strategies, increasingly opting to lease rather than build data centers, thereby transferring the massive debt burden and construction risks to developers. Microsoft CEO Satya Nadella candidly stated that predicting an "overbuild" of computing power is one of the reasons he is "happy to be a tenant." This strategic shift is concentrating risk on relatively inexperienced emerging developers.

New Players Entering: From Energy Companies to Bitcoin Miners

The participants in this AI infrastructure race come from a diverse range of backgrounds, highlighting a fundamental shift in the market.

In the southern Italian region of Apulia, Lorenzo Avello, previously involved in renewable energy projects, is planning to create a "Mediterranean AI hub" through his new company Adriatic DC. The project includes a massive campus with a power consumption of up to 1.5 gigawatts, aiming to become one of Europe's largest computing business centers. Avello himself has no prior experience in building data centers.

Similar ambitions are emerging in North America. Kevin O'Leary, host of the reality TV show "Shark Tank," plans to develop "the world's largest AI data center industrial park" in Alberta, Canada, leveraging the region's abundant natural gas and geothermal resources. He confidently states, "I will be one of the five people who successfully build a super-scale facility, and you will see hundreds fail."

At the same time, companies in the cryptocurrency sector are also making significant inroads. Bitcoin mining company Bitdeer Technologies Group plans to invest "hundreds of millions or even billions of dollars" to expand its AI cloud business. The company's Chief Strategy Officer Haris Basit stated that they plan to launch an AI data center campus with a capacity of 570 megawatts in Clarington, Ohio, in the second half of 2027. Bitdeer is not alone; former cryptocurrency mining company CoreWeave Inc. has become a leader in renting Nvidia chips to companies like Microsoft

Trillions of Capital Flowing In and Concerns About Bubbles

The influx of massive capital is accompanied by growing concerns about an AI bubble.

The Stargate infrastructure project announced this year by OpenAI, Oracle, and SoftBank Group promises to invest up to $500 billion in rural areas of the United States. The project in Italy led by Lorenzo Avello is also expected to cost over €50 billion (approximately $59 billion). However, Wall Street veteran and contrarian investor Michael Burry has warned about the AI bubble. The cyclical nature of recent transactions in the AI field—where companies like NVIDIA and Microsoft invest in startups that purchase their products—has also raised market doubts about the real demand for AI products.

“There aren’t that many customers,” said Charles Fitzgerald, a former Microsoft manager who has long focused on data centers. He believes that many of the projects being developed will never come to fruition.

Howard Marks, co-founder of Oaktree Capital Management LP, also warned that the risk of overbuilding is increasingly looming. He pointed out that savvy tech companies have included flexibility clauses in their computing power leasing contracts, which could leave data center owners in a difficult position if they exit the contracts.

Strategy Shift of Tech Giants: From Building to Leasing

Faced with huge capital expenditures and uncertain long-term demand, tech giants are cleverly shifting the risk away.

Take Meta as an example; the company secured about $60 billion in October for building data centers, but half of that ($30 billion) will not appear on its balance sheet. This deal, arranged by Morgan Stanley, places the financing in a special purpose vehicle (SPV) associated with asset management company Blue Owl Capital Inc.

Microsoft has also committed to investing over $60 billion in leasing agreements with "neoclouds," companies responsible for building data centers equipped with advanced AI chips. This strategy allows tech giants to enjoy the benefits of computing power expansion while avoiding the direct burden of construction and operational heavy asset risks, leaving the pressure on developers.

Execution Risks Highlighted: Project Delays and Contract Terminations

For new players taking on risks, the road from blueprint to reality is fraught with challenges.

Former cryptocurrency mining company CoreWeave recently lowered its annual revenue forecast due to “temporary delays related to a third-party data center developer that is behind schedule.” Bitdeer’s Bitcoin mining facility under construction in Ohio also experienced a fire last month. Data center power developer Fermi Inc. saw its stock price plummet 46% on the same day it disclosed that an investment-grade tenant had terminated a $150 million agreement.

These events expose the severe challenges faced by newcomers in project management, construction safety, and contract stability. Although Bitdeer’s Haris Basit claims that their contract terms are “rock solid” and even hopes for client defaults to collect exit fees and re-lease at higher prices, systemic risks still exist As Basit acknowledged, despite the company's diversified layout across multiple dimensions such as Bitcoin and AI, self-operated and custodial services, Europe and the United States, "you can imagine a scenario where everything is affected. You may not be able to guard against this situation." This is the core uncertainty that investors must face when evaluating the AI data center gold rush